Gadolinium borosilicate glass-bonded Gd-silicate apatite: A glass-ceramic nuclear waste form for actinides

GADOLINIUM BOROSILICATE GLASS-BONDED Gd-SILICATE APATITE: A GLASS-CERAMIC NUCLEAR WASTE FORM FOR ACTINIDES Donggao Zhao*, Liyu Li**, L.L. Davis**, W.J. Weber**, R.C. Ewing* *Department of Nuclear Engineering & Radiological Sciences, University of Michigan, Ann Arbor, Michigan 48109-2104 **Pacific Northwest National Laboratory, Richland, Washington 99352 ABSTRACT A Gd-rich crystalline phase precipitated in a sodium gadolinium alumino-borosilicate glass during synthesis. The glass has a chemical composition of 45.4-31.1 wt % Gd2O3, 28.8-34.0 wt % SiO 2, 10.8-14.0 wt % Na2O, 4.3-5.9 wt % Al2O3, and 10.8-14.9 wt % B2O3. Backscattered electron images revealed that the crystals are hexagonal, elongated, acicular, prismatic, skeletal or dendritic, tens of µm in size, some reaching 200 µm in length. Electron microprobe analysis confirmed that the crystals are chemically homogeneous and have a formula of NaGd9(SiO4)6O2 with minor B substitution for Si. The X-ray diffraction pattern of this phase is similar to that of lithium gadolinium silicate apatite. Thus, this hexagonal phase is a rare earth silicate with the apatite structure. We suggest that this Gd-silicate apatite in a Gdborosilicate glass is a potential glass-ceramic nuclear waste form for actinide disposition. Am, Cm and other actinides can easily occupy the Gd-sites. The potential advantages of this glassceramic waste form include: 1) both the glass and apatite can be used to immobilize actinides, 2) silicate apatite is thermodynamically more stable than the glass, 3) borosilicate glass-bonded Gd-silicate apatite is easily fabricated, and 4) the Gd is an effective neutron absorber. INTRODUCTION Considerable effort has been devoted to determining distributions and solubility limits of radionuclides and neutron absorbers in borosilicate glasses in the four-component Na2O-B2O3Al2O3-SiO2 system [1-3]. In recent years, glass compositions in this system have been developed to test the effects of changing compositions on the solubilities of the neutron absorbers Gd and Hf [1-3]. When Gd exceeds its solubility limit in the glass during synthesis of Gd-borosilicate glasses, a Gd-rich crystalline silicate phase precipitates. Likewise, Gd-rich crystalline silicate phases precipitate in Gd-borosilicate glasses during extended times at elevated temperatures [4], as would occur during slow cooling of canisters or during repository storage. The purpose for synthesizing a Gd-borosilicate glass is to incorporate as much Gd as possible into the glass. Therefore, the precipitation of a Gd-rich crystalline phase from the glass matrix was initially considered unwanted. However, as a Gd-rich phase, this crystalline phase may also immobilize actinides, as has been demonstrated previously [4]. An essential issue of the long-term immobilization and disposal of actinides is whether a waste form is sufficiently durable. Durability may be based on a variety of chemical and physical properties, e.g., mechanical strength, thermodynamic stability, slow kinetics of corrosion, or low diffusivity of radionuclides and neutron absorbers. Durable phases that survive weathering, erosion, transport, and deposition over geologic periods are potential candidates for actinide waste forms. Such phases include zircon, ZrSiO 4, calcium phosphate apatite, Ca4-xREE6+x(SiO4)6y (PO4)y (O,F)2, monazite, (Ce,La,Nd,Th)PO4, pyrochlore, (Ca,REE)2Ti2O6(OH,F), zirconolite, (Ca,Th,Ce)Zr(Ti,Nb)2O7, baddeleyite, ZrO2, and zirconia, ZrO2 [1, 4, 5]. In this paper, we characterize the Gd-rich crystalline silicate phase in terms of its chemistry, structure and susceptibility to radiation damage and propose Gd-silicate in a Gd-borosilicate glass matrix as a possible glass-ceramic nuclear waste form for actinides. A variety of other 1 of 8 D. Zhao

Gd-bearing silicates have been reported previously. These Gd-silicates include digadolinium silicate oxide Gd2SiO5 [6], gadolinium disilicate Gd2Si2O7 [7-8], gadolinium silicate oxide Gd14Si9O39 [9], lithium gadolinium oxide silicate LiGd9Si6O26 [10], sodium gadolinium silicate NaGdSiO 4 [11], sodium gadolinium silicate hydroxide NaGdSiO 4(NaOH)0.213 [12], sodium gadolinium tecto-alumosilicate hydrate Na7Gd27(Al88.11Si103.9O384)(H2O)195 [13], and sodium gadolinium silicate fluoride oxide (Na1.19Gd8.81)(SiO 4)6(F0.38O1.62) [14]. EXPERIMENT Sample Synthesis Sodium gadolinium silicate crystals precipitate from borosilicate glasses when Gd exceeds its solubility limits [1, 3]. The baseline borosilicate glass has a composition of 15B2O3⋅20Na2O⋅5Al2O3⋅60SiO 2 (mole %). Baseline glasses were synthesized in a covered Pt 10 % Rh crucible at temperatures between 1110°C and 1400°C from well-mixed powders of SiO2, Al2O3, H3BO3 and Na2CO3. The borosilicate glasses were then quenched and ground to approximately 200 mesh. The glasses were melted again at 1450°C after adding Gd2O3 [1]. The glass melts, with varying amounts of Gd2O3, were quenched by immersion of the base of the crucible in water. The resulting products are transparent, colorless sodium gadolinium borosilicate glasses; one of these glasses (B15Gd48) contained an exotic crystalline phase, a sodium gadolinium silicate apatite. Analytical techniques The sodium gadolinium alumino-borosilicate glass and the precipitated gadolinium silicate apatite were studied by electron microprobe analysis (EMPA), scanning electron microscopy (SEM), and X-ray powder diffraction (XRD). The EMPA and SEM work was done in the Electron Microbeam Analysis Laboratory at University of Michigan. The XRD work was done at Pacific Northwest National Laboratory and University of Michigan. Backscattered electron (BSE) images and energy dispersive X-ray spectra (EDS) were used to qualitatively identify elements in the phases. A four-spectrometer Cameca CAMEBAX electron microprobe analyzer was used to obtain chemical compositions of the glass and the crystalline precipitates. Powder X-ray diffraction was used to determine the crystal structure of the precipitated silicate apatite. The Cameca PAP correction routine φ(ρz), i.e., modified ZAF (atomic number Z, Absorption and Fluorescence) by Pichou and Pichouir [15], was used in data reduction. EMPA procedures used are given in reference [16-17]. The accelerating voltage was 20 kV; the beam current was 15 nA; and the peak and background counting times were 30 and 15 seconds, respectively. Large beam sizes, lower beam current, and a shorter counting time were used to avoid possible Na migration. RESULTS AND DISCUSSION Gadolinium borosilicate glass The measured chemical compositions of the crystal-bearing sodium gadolinium aluminoborosilicate glass by EMPA are 45.39-31.13 wt % Gd2O3, 28.80-34.04 wt % SiO 2, 10.7514.02 wt % Na2O, 4.30-5.89 wt % Al2O3, and 10.75-14.91 wt % B2O3 (Table I), which are heterogeneous [16]. There are two compositional domains identified in the glass host. In the BSE image (Figure 1A), the glass matrix from the darker upper left area is enriched in Si, Al and Na and depleted in Gd; whereas, the glass matrix from the brighter region is enriched in Gd (Table I). The heterogeneity of the glass matrix may be the result of crystallization that 2 of 8 D. Zhao

removes Gd from the glass matrix. The differences between the target and the measured compositions are significant. The glass matrix has 31.13 wt % Gd2O3 in the darker areas and 45.39 wt % Gd2O3 in the brighter areas, both of these compositions differ from the target composition of 48.00 wt % Gd2O3. The composition of the brighter areas is closest to the target composition (Table I). Table I. Target and measured compositions (wt %) of the glass host and measured composition of the precipitated Gd-silicate apatite in sample B15Gd48 target Average

bright glass domain

dark glass domain

crystal

5 analyses

4 analyses

22 analyses

SiO2

29.30

28.8±0.3

34.0±0.3

15.7±0.2

Al2O3

4.14

4.3±0.1

5.9±0.1

0.0

Na2O

10.07

10.8±0.2

14.0±0.3

1.4±0.1

Gd2O3

48.00

45.4±1.4

31.1±1.1

81.6±2.0

B2O3 8.49 10.8 14.9 1.7 Note: 1 σ analytical errors from counting statistics given for measured compositions; B2O3 by difference.

Figure 1. Backscattered electron images of the glass containing the precipitated crystalline phase in sample B15Gd48. A) Elongated, acicular, prismatic or dendritic Gd-silicate apatite crystals; cross sections are often hexagonal; glass matrix in the upper left area slightly darker than those in the central and lower right areas, indicating chemical heterogeneity of the matrix. B) Enlarged image of the central area in A. The precipitated crystals are on average tens of µm in size, but some are up to 200 µm in length. 3 of 8 D. Zhao

Gd-silicate apatite The Gd-silicate crystals that precipitated from the sodium gadolinium alumino-borosilicate glass are elongated, acicular, skeletal, prismatic or dendritic; and most crystals are tens of µm in size, but some crystals are up to 200 µm in length (Figure 1B). Cross sections of the crystals are often hexagonal, sometimes skeletal with hexagonal euhedral voids at their centers. The composition of the crystals was obtained from EMPA and tabulated in Table I. The electron microprobe analyses obtained during three analytical sessions were the same, although they were obtained with different beam sizes. Na2O contents of the precipitated crystals were the same when analyzed using different electron beam sizes (from a point beam to a 15 × 15 µm rastered beam), indicating that alkali, i.e., sodium, loss is not significant. The mean and standard deviations of the 22 analyses are: SiO 2 15.66 ± 0.30 wt %, Gd2O3 81.25 ± 0.81 wt %, and Na2O 1.38 ± 0.06 wt %. The crystals of different shapes have the same composition, and each crystal is chemically homogeneous. The chemical formula of this phase based on EMPA is approximately NaGd9(SiO4)6O2 or NaGd9Si6O26, which has the same formula as LiGd9Si6O26 [10]. Some minor B may substitute for Si in the structure [16]. The powder X-ray diffraction pattern of the precipitated crystals can be indexed in the hexagonal system and is similar to that of lithium gadolinium silicate LiGd9Si6O26, a rare earth silicate apatite. The latter has space group P63/m or P63, and cell parameters a = 0.9407 nm and c = 0.6842 nm based on powder and single-crystal X-ray examination [10]. The chemical and structural examinations demonstrate that the precipitated crystalline phase is a rare earth silicate apatite [17-18]. These crystalline phases belong to the mixed-cation oxyapatite group [19-21]. The formula of apatite, Ca4Ca6(PO4)6(F,Cl,OH)2, can also be expressed as A(REE)9(SiO4)6O2, where A (4f site), which can be Li or Na, is coordinated by nine silicate oxygens, and REE (three 4f and six 6h sites) are 4f-transition lanthanides in nine-fold (4f site) or seven-fold (6h site) coordination. If P substitutes Si in silicate apatite, the formula can also be rewritten as A4-x(REE)6+x(SiO4)6-y (PO4)y (F,OH,O)2, where A could also be Mg, Ca, Sr, Ba, Pb and Cd. Radiation stability of silicate apatite One of the primary concerns with crystalline Gd-silicate apatite as an actinide waste form is the radiation stability of the phase, i.e., its resistance to radiation-induced amorphization due to α-decay of actinides [22]. Radiation damage of nuclear waste forms can result in significant changes in volume, leach rate, stored energy, structure and mechanical properties [23-24]. Although radiation damage effects on this specific type of silicate apatite NaGd9(SiO4)6O2 have not yet been studied, a considerable number of studies are already available on radiation effects on other rare earth silicate apatites such as Ca3Gd7(SiO4)5(PO4)O2 [4, 25-26], Ca2Nd8(SiO4)6O2 [27-28], Ca2La8(SiO4)6O2 [29-31], and Ca5(PO4)3(F,OH,Cl) [32]. In a study of partially devitrified Gd-borosilicate glass containing 244Cm, Weber et al. [4] demonstrated that the silicate apatite, Ca3Gd7(SiO4)5(PO4)O2, formed as either small (